Product
A new era for forensic watermarking
STARDUSTmark closes production security gaps, enables content ownership tracing, and detects tampering. Media is protected by embedding identification data imperceptible to the untrained eye – with remarkable extraction precision from just one image or photo of a video playing.
Deploy Hollywood-recognized forensic watermarking across a multitude of video, still image, and document use cases with one technology solution.
Tighten your content workflow security by leveraging modern opportunities to track leaks that previously weren’t possible.
Watermarking is more accessible than ever with affordable protection to match your use case through our flexible, low-cost business model.
Keep provenance and ownership metadata intact, or gain authenticity confidence through tamper and deepfake detection.
Our steganography approach applies hidden marks – just as imperceptible as traditional solutions based on perceptual metrics.
Key facts
Robust extraction with one image
STARDUSTmark offers highly resilient extraction – resisting a wide range of distortions and attacks from just a single image or video frame.
Single-frame watermarking
For video, we watermark every frame letting you track clips, screenshots, or even a snapped photo.
Distortion resistance
Our extraction accommodates mirroring and diverse geometric distortions when images are captured from different angles, positions, and distances.
Filtering tolerance
Our mark withstands color conversion, sharpening, and many other image alterations.
Obstruction flexibility
Other solutions mark only portions of content where obstruction or cropping can defeat it. We protect entire assets enabling extraction even when partially visible.
Downscale endurance
Dramatically downscaled content can still be identified.
Compression durability
Our watermark survives low bitrate encoding for video and single-image files.
Product
Easy embedding options
Either server-side in the cloud, on-prem, or client-side, we’ll integrate into your deployment structure.
Take advantage of easy embedding through our powerful Video Toolkit and Content Platform cloud content processing services or use plugins for industry-leading tools like AVID Media Composer, Dolby Hybrik, and FFmpeg.
PROJECT
We helped Iyuno, a global leader in media localization, deploy STARDUSTmark across its security workflows.
Try STARDUSTmark today
Get started now to discover how we can secure your valuable digital media.
Memberships
Industry members
We’re members of leading industry and standardization bodies to help address studio security challenges and advance media authenticity and provenance frameworks.
FEATURES
Your forensic watermarking features
Discover what advanced abilities are available to you.
See what’s new in STARDUSTmark version 3!
Single-frame & image forensic watermarking
| Protect video, images, and documents |  |
| Watermark each entire frame | |
| Imperceptible mark Based on VMAF, SSIM, SSIMULACRA2, PSNR-HVS metrics and desired robustness level. | |
Ability to layer many keyed watermarks on top of each other. For example: having a distributor mark and session mark at the same time. | |
| Confidence level reporting | |
Contact us for example metric setups, scales, and score values. | Very high to excellent |
| Unique payload IDs | Trillions + |
| Tamper & deepfake detection With heatmapping for tampered areas. | |
| Codec, container, and streaming format agnostic | |
Extraction
Watermark robustness can be tuned to match your use-case. For example: You can set extraction to survive camcording or set it to maintain the highest-quality visuals. | |
| High resilience from compression, scaling, distortion, cropping, and filtering | |
| Safeguards against screencasting and screenshots | |
Resists both alternating different watermarked sources over time, as well as combining frame/image pieces from different sources. | |
| Withstands HDCP stripping | |
| Survives camcording | |
| Works through digital/analog conversion | |
For watermarked images, extraction can be accomplished in as little as a few seconds. | As fast as a few seconds |
| Blind extraction Source content isn’t needed for watermark extraction. | |
| Non-blind extraction Source content is used for extraction for higher robustness. | |
Embedding
| Real-time embedding | ~1 ms / frame or image (depending on hardware) |
| Server-side (cloud or on-prem) | |
| Client-side Using an alpha blend overlay. | |
Our algorithm finds the best areas in a frame or image to place a watermark where it’s the most imperceptible while preserving robustness. For example: Placing a watermark on a mountain side vs. a clear blue sky. | |
| No profiling required | |
| Library availability | FFmpeg or C |
| AVID Media Composer® plugin | |
| Dolby Hybrik plugin | |
Added directly into media processing pipelines to simplify embedder deployments. | |
| Pre-integrated with our WebRTC realtime streaming solution | |
| Pre-integrated with our remote desktop solution | |
| Pre-integrated with our Video Toolkit service | |
| Pre-integrated with our Content Platform service | |
Popular use-cases
| Production & post-production | Proxy-based workflows including: dailies, editing, localization, review/approval, designs, storyboards |
| Pre-release workflows | Including screener, trailer, poster distribution |
| Real-time WebRTC streaming | |
| Remote desktop & collaboration workflows | |
FAQs
Have any questions?
Here are some common questions we get about STARDUSTmark, but feel free to get in touch if you haven’t found what you’re looking for.
We use these video quality metrics to quantify the visual impact of embedding a forensic watermark. STARDUSTmark aims to remain robust for detection while keeping degradation as low as possible across these metrics, ensuring our watermark is effectively imperceptible to human viewers.
- SSIM (structural similarity index measure) measures similarity between two images based on structure, luminance, and contrast. It’s widely used and easy to interpret.
- SSIMULACRA2 is a more advanced, perceptually tuned metric developed to better reflect human visual sensitivity. It captures artifacts (e.g. ringing, banding) that SSIM may overlook, making it useful for subtle distortions.
- PSNR-HVS (peak signal-to-noise ratio – human visual system) extends traditional PSNR by incorporating characteristics of human vision, such as reduced sensitivity to high-frequency noise. It provides a better indication of perceived quality than plain PSNR.
- VMAF (video multi-method assessment fusion), developed by Netflix, combines several quality metrics using machine learning to evaluate video quality. However, it’s mainly designed to assess encoding and compression artifacts, so has limited use for characterizing distortions caused by watermarking.
A more appropriate solution would be A/B watermarking which can scale efficiently for large audience numbers. Learn about our other forensic watermarking options through our Video Toolkit and Content Platform solutions.